VELA home
Technology

Evidence from more than one signal.

Pressure trends reveal overall change. Local structural signals add spatial information. VELA is designed to bring them together.

CAD exploded view of the acquisition unit
System architecture

From acquisition to insight.

BUILT / Acquisition & logging
SENSOR INPUTS
Pressure node0–5 V · ADS1115
PVDFCharge amplifier · AD7606
CONTROLLER
ESP32-S3Time-stamped acquisition
OUTPUT MODES
USB → Live dashboardConnected operation
MicroSD → CSVOffline recording
IN DEVELOPMENT / Area prioritization
Baseline comparisonCompare each channel with its own history
Multi-point fusionPressure, signal timing and energy
Priority rankingRegion, evidence and action
Operating concept

An operating loop, not just an alarm

The intended workflow links routine monitoring to inspection. Acquisition and logging are working in the prototype; adaptive capture and area ranking are in development.

  1. 01

    Install

    Place fixed sensors and map their positions before service.

  2. 02

    Establish a baseline

    Record each channel under known normal conditions.

  3. 03

    Monitor

    Track pressure and local signals on a shared timeline; selective higher-rate sensing is planned.

  4. 04

    Rank

    Compare channels with their own baselines, then combine nearby responses and repeated evidence.

  5. 05

    Inspect & log

    Recommend a region for local confirmation; record the inspection result and update the baseline when appropriate.

Three design principles

Intelligence at the system level.

S1S2S3S4AcquisitionPVDF responsePressureConcept illustration · not measured data

Multi-modal fusion

Pressure fluctuation and localized structural response offer complementary information.

Signal interpretation

Amplitude alone is not a location.

A channel’s mounting and background vibration affect its normal signal. Per-channel comparison is central to VELA’s proposed analysis.

Every sensor has its own normal

A sensor near a pump may be louder than one near a leak. Comparing absolute amplitudes alone risks measuring the installation rather than the event.

Compare changes, not raw levels

Use each sensor’s healthy record as a reference for features such as RMS and energy. Combine pressure conditions, nearby responses and repeated trials to support a region ranking.

Absolute level (arbitrary units)

BaselineCurrent
S1
100 → 104
S2
40 → 55.2
S3
70 → 74.9
S4
50 → 51

Change from own baseline

Change from own baseline
S1
+4%
S2
+38%
S3
+7%
S4
+2%

S2 has the largest relative change—not a confirmed leak location.

Illustrative values, not measured data.
Monitoring interface

Four states. Clear actions.

VELA monitoring console simulation

This is the monitoring interface’s decision model; the interactive console demonstrates the intended actions.

Try the interactive console

Normal

Evidence

Channels remain within their expected ranges; data is complete.

Recommended action

Continue routine monitoring.

Watch

Evidence

A change is present, but the region order is not stable.

Recommended action

Observe, collect more data or check operating conditions.

Inspect

Evidence

A consistent region lead has formed in the intended workflow.

Recommended action

Prioritize a local check and record the outcome.

Unknown

Evidence

Channels are offline, data is incomplete or evidence conflicts.

Recommended action

Check channel health or collect a complete window before ranking.

Integrated hardware
Engineering focus

The value is in the decision logic.

VELA’s development focuses on calibration, sensor selection, synchronization and evidence fusion. The architecture is designed for modular expansion across pressurized structures.